../

a16z

What the people behind Andreessen Horowitz (a16z) actually wrote about markets, product/market fit, running a company through bad times, and network effects, with the year and source for each idea and where it applies in the idea-to-MVP playbook. Compare with Y Combinator, which starts from users rather than markets, and Peter Thiel on monopoly.

Who they are and why listen

WhoBackgroundWhy listenMain sources
Marc Andreessenco-created the Mosaic browser, co-founded Netscape, Loudcloud/Opsware and a16z (2009)a founder who went through the 1990s boom and bust, then a VCThe Pmarca Guide to Startups (2007 blog), essays
Ben HorowitzNetscape product manager, then co-founder and CEO of Loudcloud/Opsware (sold to HP in 2007), a16z co-founderthe most candid writer on what being a CEO in trouble feels likeThe Hard Thing About Hard Things (2014), What You Do Is Who You Are (2019), a16z blog
Chris Dixonfounder (Hunch, SiteAdvisor), a16z general partnershort, sharp essays on how new technologies startcdixon.org
Andrew Chengrowth writer, Uber growth executive, a16z general partnerthe network-effects playbookThe Cold Start Problem (2021), andrewchen.com
Benedict Evansformer a16z partner, now independent analystbig-picture tech strategyben-evans.com newsletter and presentations

Where YC's advice is "talk to users", a16z's center of gravity is the market: pick a market that pulls the product out of you, then fight to reach product/market fit.

Product/market fit: the only thing that matters

Andreessen's Part 4: The only thing that matters (opens in a new tab) (June 2007) is the essay that popularised the term product/market fit (PMF).

He splits a startup into team, product and market, and says most entrepreneurs and VCs name team, most engineers name product, and he takes the third position: market matters most. His reasoning, paraphrased:

  • In a great market, "the market pulls product out of the startup": the product just has to basically work, and the team can be upgraded on the fly.
  • In a terrible market, the best product and team still fail.
  • Great products can create huge new markets (his example is VMware), but he wouldn't count on it.

He credits Andy Rachleff (Benchmark) for the formulation, which he calls Rachleff's Law of Startup Success:

"The #1 company-killer is lack of market."

Rachleff's lawOutcome
great team + lousy marketmarket wins
lousy team + great marketmarket wins
great team + great market"something special happens"

And Rachleff's Corollary: "The only thing that matters is getting to product/market fit." His definition: "Product/market fit means being in a good market with a product that can satisfy that market."

When PMF isn't happeningWhen PMF is happening
customers aren't quite getting valuecustomers buy as fast as you can make it
word of mouth isn't spreadingusage grows as fast as you can add servers
usage isn't growing that fastmoney piles up in the bank account
press reviews are "blah"you hire sales and support as fast as you can
sales cycles take too long; deals don't closereporters call you

Before PMF (BPMF) do whatever it takes to get there: change people, rewrite the product, change market, say yes or no to customers against your instincts, raise that dilutive round. After PMF, everything else starts to matter. He adds that founders of successful startups usually credit things that had nothing to do with it: "People are terrible at understanding causation."

More from the Pmarca guide

The Pmarca Guide to Startups (opens in a new tab) was a 2007 blog series, now archived at pmarchive.com.

PrincipleWhat it meansSourceApply it by…
The onion theory of riskinvestors see a startup's risk as layers; your job is to peel them off one by one until investing looks merely risky, not terrifyingPart 2: When the VCs say "no" (opens in a new tab) (2007)listing your risks and naming the one the next six weeks removes
Kill market risk with customersdetailed research sometimes helps, but usually you need actual (preferably paying) customerssamesmoke tests and pre-sales
Kill product risk by buildingfor technology and product risk the only answer is to build it, at least to betasamean MVP that tests the riskiest assumption
Big companies are Moby Dickyou are Captain Ahab: you can't predict or explain what a big company will do, so don't build your plan around onePart 5 (opens in a new tab) (2007)not depending on a partnership or acquirer
How much to raisebefore PMF, raise at least enough to reach PMF; after PMF, enough to fully exploit the opportunity and reach profitabilityPart 6 (opens in a new tab) (2007)sizing a round to a milestone, not a vanity number
The plan doesn't matter muchyou won't know if the idea works; expect to change every part of the planPart 7 (2007)treating the canvas as a hypothesis

The onion's layers in Part 2:

RiskQuestionHow to peel it
founderis this the right team?add or swap founders
marketwill anyone want it and pay?real customers
competitionare others doing it?a differentiated angle
timingtoo early or too late?evidence the market is ready now
financingcan it raise the later rounds it needs?a plan that needs less capital
marketingcan it cut through the noise?a channel that works
distributioncan it reach the customer?a partner or direct channel
technologycan it be built at all?build it
productcan this team build it?build it
hiringcan it hire who it needs?name the people
locationis it where it can hire and sell?move, or show why it doesn't matter

Big theses: software, building, optimism

These are worldview essays, not tactics, but they explain what a16z funds.

EssayYearCore claimFor a founder it means…
Why Software Is Eating the World (opens in a new tab)2011 (Wall Street Journal)"Software is eating the world": ever more industries are run on software, and software companies will disrupt incumbentslook for old industries where software now works at scale; the "why now?" test
It's Time to Build (opens in a new tab)2020written early in COVID-19: the West lacked masks, tests and housing because "We chose not to build"; the problem is desire, inertia, regulatory capture and willambitious "atoms" problems are fundable
The Techno-Optimist Manifesto (opens in a new tab)2023a statement of beliefs (e.g. growth is progress; technology is a lever on the world)context for a16z's positions; it drew wide praise and criticism

Ben Horowitz: the hard things

The Hard Thing About Hard Things (Harper Business, 2014) grew out of Horowitz's a16z blog posts, cited here where the text is online.

The Struggle (2012)

The Struggle (opens in a new tab) describes the stretch when nothing works, in a run of "The Struggle is…" lines ("when food loses its taste", "when you are surrounded by people and you are all alone"). His points: every great entrepreneur has been through it; it is not failure, but it causes failure if you're weak; and "The Struggle is where greatness comes from."

Nobody Cares (2011 (opens in a new tab)): after an injury crisis, the football coach Bill Parcells was told by Al Davis, "Bill, nobody cares, just coach your team." Horowitz's reading: spend no energy on what you could have done and all of it on the way out.

Peacetime CEO / Wartime CEO (2011)

Peacetime CEO/Wartime CEO (opens in a new tab): peacetime is when the company has a large advantage in a growing market; wartime is when it faces "an imminent existential threat". He says he was a peacetime CEO for about nine months and a wartime CEO for the next seven years, and that most management books describe peacetime techniques only.

Peacetime CEOWartime CEO
knows that proper protocol leads to winningviolates protocol in order to win
focuses on the big picture; empowers people to make detailed decisionscares about any detail that interferes with the prime directive
spends time defining the culturelets the war define the culture
works to minimize conflictheightens the contradictions
strives for broad-based buy-inneither indulges consensus-building nor tolerates disagreements
aims to expand the marketaims to win the market

(Rows lightly condensed from his list of about 20.) An MVP-stage startup is almost always in wartime, which is why founder-mode behavior feels natural early on (see PG's Founder Mode).

Other Horowitz rules

PrincipleWhat it meansSourceApply it by…
Lead bulletswhen you're losing because the product is worse, no clever strategy saves you; an engineer told him "There is no silver bullet that's going to fix that. No, we are going to have to use a lot of lead bullets."Lead Bullets (opens in a new tab) (2011)fixing the product before pivoting when customers are buying, just not from you
People, products, profits, in that ordera chapter title and rule of the book: look after people first, because products and profits followThe Hard Thing (2014)early culture and hiring
A good PM is the CEO of the productan old Netscape-era training memo contrasting good and bad product managers (e.g. good ones define the "what", bad ones obsess over the "how"; good ones take written positions)Good Product Manager/Bad Product Manager (opens in a new tab) (reposted 2012; it says it was written about 15 years earlier)writing crisp specs for the MVP
Culture is what you doculture is how a company makes decisions: the assumptions people use for everyday choices, set by behavior, not value statements; ignore something off-culture and you've created a new cultureWhat You Do Is Who You Are (opens in a new tab) (2019; publisher summary)deciding early which behaviors you will enforce, then enforcing them

Chris Dixon: toys, tools and mazes

IdeaWhat it meansSourceApply it by…
The next big thing starts out looking like a toybuilding on Christensen: disruptive products undershoot user needs at first, so incumbents dismiss them; judge them as processes riding external improvement curves (cheaper chips, more bandwidth), not as static productsThe next big thing will start out looking like a toy (opens in a new tab) (2010, written before he joined a16z)asking what curve will make your "toy" good enough, and when
Most toys stay toysDixon says so himself; sustaining products can still be good businessessamenot using "it looks like a toy" as evidence
Come for the tool, stay for the networkattract users with a single-player tool, then get them into a network (Delicious, Instagram)Come for the tool, stay for the network (opens in a new tab) (2015)networked products that are useful with one user
The idea mazeBalaji Srinivasan's term: a good idea is a multi-year map of which turns lead to treasure and which to death; build it from history, analogy, theory and direct experienceThe idea maze (opens in a new tab) (2013)a history of prior attempts in competition analysis
Stealth mode is a bad idealearning the maze from practitioners outweighs the risk of your idea being stolensametalking about your idea openly

Andrew Chen: networks and channels

The Cold Start Problem (2021)

Chen's book (book site (opens in a new tab)) is organized as five stages, each a part of the book:

StageProblemKey ideas (chapter topics)Apply it by…
1. The cold start problema network with too few users is useless, so users leave (anti-network effects)the atomic network (the smallest network that is stable and can grow by itself); the hard side (the minority who do most of the work: hosts, drivers, creators); a killer product; magic momentsdesigning the MVP around one atomic network
2. Tipping pointreplicating atomic networks until the market tipsinvite-only launches; "come for the tool"; paying up for launch; "flintstoning" (humans doing the work software will do later); hustlefirst users in launch
3. Escape velocitymaking growth self-sustainingthe trio of forces: engagement, acquisition and economic effectsgrowth loops
4. The ceilinggrowth stallssaturation, the law of shitty clickthroughs, the network revolting, "eternal September", overcrowdingwatching cohort quality as you grow
5. The moatdefending against other networksvirtuous and vicious cycles; cherry-picking; big-bang launch failures; competing over the hard side; bundlinglong term, after PMF

"Flintstoning" is Chen's version of PG's "be the software" in Do Things That Don't Scale (opens in a new tab).

The Law of Shitty Clickthroughs (2012)

The Law of Shitty Clickthroughs (opens in a new tab) (written before he joined a16z):

"Over time, all marketing strategies result in shitty clickthrough rates."

His example: the first banner ad (HotWired, 1994) had a 78% clickthrough rate; Facebook banner ads in 2011 averaged 0.05%. Causes: novelty wears off, competitors copy what works, and early adopters respond better than the mainstream. The practical rule is that a channel that works today is decaying; keep testing new ones and budget for falling performance. Use it in channels.

Benedict Evans, briefly

Evans (a16z partner until 2019, now independent) writes analysis rather than founder advice: what platform shifts mean, which markets are big, why incumbents miss changes. For an early founder his use is market context for the "why now?" question. See his essays and presentations (opens in a new tab).

Where each piece applies

Pipeline phasea16z ideaWhere
Ideationmarket first; software eating an industry; the idea maze; toy vs processideation
Validationpeel market risk with real customers; onion of risksvalidation
Product designatomic network; hard side; come for the toolproduct design
Buildinglead bullets: fix the product before hunting for tricksbuilding the MVP
Launch & iteratePMF signs; decaying channels; tipping pointlaunch & iterate
Whole journeythe Struggle; wartime operating; nobody caresplaybook

Critiques and limits

  • Hindsight on PMF. "You can always feel it" is only obvious afterward. In the middle, founders see mixed signals: use measurable proxies (retention cohorts, the Sean Ellis test, organic referrals) from the PMF section, not a feeling.
  • Market-first vs founder-first. Andreessen says market beats team; YC and many VCs say they bet on founders. Both are partly true: a great team can find a better market (pivots), but no team can fix a market that doesn't exist.
  • Survivorship. "Looks like a toy" and "wartime CEO" stories are told about winners. Many toys stayed toys; many wartime CEOs lost the war.
  • Wartime as an excuse. Horowitz's wartime list (profanity, intolerance of disagreement) can be used to justify bad management in companies that aren't at war. The honest test is whether there really is an existential threat.
  • Network effects are rare. Altman notes few businesses have true network effects (How to Be Successful (opens in a new tab), 2019). Don't force a network into a tool that doesn't need one.
  • Politics and positioning. a16z publishes openly political manifestos and lobbies on policy (e.g. crypto and AI). That's context for why some readers discount the firm; judge each argument on its evidence.
  • Era. The Pmarca guide is from 2007, before cloud computing made building cheap; "raise enough to reach PMF" now often means less money than it did then.

Takeaways checklist

MARKET
[ ] Is the market big and growing? Evidence, not hope
[ ] Would the market pull a basic working product out of
    us? Who is already trying to buy something like it?
[ ] Why now: what curve (cost, speed, adoption) makes
    this possible or soon possible?
[ ] Idea maze drawn: who tried before, why they died
 
RISK
[ ] Top 3 onion layers listed; next milestone peels one
[ ] Market risk tested with real (ideally paying) users
 
PMF
[ ] Before PMF: everything else is secondary
[ ] PMF judged on retention and pull, not feelings
[ ] Round sized to reach PMF, not a round number
 
NETWORKS (if any)
[ ] Atomic network defined; hard side identified
[ ] Useful with a single user (tool before network)?
[ ] Channel decay expected; new channels in test
 
OPERATING
[ ] Losing on product? Lead bullets, not a pivot
[ ] Wartime or peacetime? Honest answer
[ ] Culture rules decided by actions, not a poster

References